Skip to main content
AITier

Qwen3.8 Max (0902)

AlibabaQwenOpen WeightQwen3.8-Max License · Commercial OK

Description

Qwen3.8 Max, released as the Qwen3.8-2.4T-A95B open-weight checkpoint, is Qwen's flagship mixture-of-experts model for coding, research, professional workflows, and long-horizon agents. It has 2.4 trillion total parameters with 95 billion active parameters, a native 262,144-token context window extensible to about 1 million tokens, and always-on thinking. QwenCloud serves the same model in the Qwen3.8 Max API family with text and image input.

Release Date
2026-09-02
Parameters
2.4T
Context Length
1.0M
Modalities
image, pdf, text, video

Capability Radar

45
general
72
coding
93
reasoning
69
science
60
agents
90
multimodal

Rankings

Domain#RankScoreSource
Agentic Capability18
61.0
LS
Code Ranking42
92.0
AA
General Ranking35
78.0
AA
Math Reasoning31
91.0
LB
Multimodal Ranking1
84.0
LS
Reasoning24
88.0
LB
Science61
80.0
AA

Benchmark Scores (LLM Stats)

(LLM Stats (zeroeval))

Agents

QwenSVG1713.00 / 2000SR
AndroidWorld85.3%SR
MobileWorld77.8%SR
AndroidBench75.1%SR
CoWorkBench74.8%SR
Toolathlon72.5%SR
Workspace Bench67.7%SR
Job Bench53.4%SR
OneMillion Bench52.5%SR
Agents' Last Exam52.4%SR
MLS-Bench Lite41.0%SR
AutomationBench27.3%SR

Code

QwenReactBench1724.00 / 2000SR
PaperBench93.0%SR
QwenSWEBench80.7%SR
FrontierSWE73.5%SR
SkillsBench70.2%SR
Vision2Web69.0%SR
QwenQoderBench58.4%SR
DeepSWE 1.156.6%SR
NL2Repo55.9%SR

Healthcare

HealthBench60.2%SR

Instruction Following

IFBench82.8%SR

Knowledge

PLawBench73.2%SR
PRBench-Finance58.3%SR
PRBench-Legal57.6%SR

Long Context

MRCR v2 (8-needle)92.9%SR
LongBench v266.3%SR

Multimodal

OSWorld-Verified86.1%SR

Reasoning

GPQANYU + Cohere + Anthropic (2023)92.6%SR
Terminal-Bench 2.186.6%SR
SWE-Bench ProPrinceton NLP (2024)67.7%SR
Humanity's Last Exam (with tools, text-only)56.2%SR
Humanity's Last Exam43.6%SR

Search

WideSearch81.9%SR

Vision

VideoMME w sub.90.4%SR
RealWorldQA88.0%SR
ScreenSpot Pro84.5%SR
MMMU-Pro82.3%SR
LVBench81.8%SR
ERQA77.8%SR
PerceptionBench63.5%SR

AA Evaluation Indices

(Artificial Analysis)
Gpqa(NYU + Cohere + Anthropic (2023))
92.8
Terminalbench V2 1
88.8
Lcr(Artificial Analysis)
80.3
Coding Index(Artificial Analysis)
76.2
Scicode(UIUC + Argonne National Lab (2024))
52.1
Tau Banking
47.8
Intelligence Index(Artificial Analysis)
45.4
Hle(Center for AI Safety + Scale AI (2025))
43.1
Terminalbench V4 0
38.9

LLM Stats Category Scores

(LLM Stats (zeroeval))
Physics
90
Biology
90
Chemistry
90
Video
90
Instruction Following
80
Multimodal
80
Search
80
Spatial Reasoning
80
Grounding
80
Vision
80
Long Context
70
Reasoning
70
Structured Output
70
General
70
Agents
70
Code
70
Productivity
60
Healthcare
60
Tool Calling
60
Math
50

Pricing

Input Price$2 / 1M tokens
Output Price$6 / 1M tokens
Blended Price (3:1)$3 / 1M tokens
Cache Read Price$0.25 / 1M tokens
Cache Write Price$2.5 / 1M tokens

Speed

Tokens/sec40.0
Time to First Token1.84s
Time to Answer51.88s

Provider Price Ranking

Provider Price Ranking

6 providers

Cheapest: DeepInfraMost Expensive: EmpirioLabs AI
ProviderInputOutput
1DeepInfraCheapest
$0
$0
2Novita
$0
$0.00001
3Fireworks
$0
$0.00001
4Together
$0
$0.00001
5AlibabaPRIMARY
$2
$6
6EmpirioLabs AI
$2
$6

Compare pricing across different API providers for this model.

External Sources

LLM StatsArtificial Analysis